RU2737465C1 - Audio encoding device, an audio encoding method, an audio encoding program, an audio decoding device, an audio decoding method and an audio decoding program - Google Patents

Abstract

FIELD: data processing.SUBSTANCE: invention relates to encoding audio signals. Technical result is achieved by encoding side information, where side information is calculated from a predictive signal for calculating the predicted value of the audio parameter in order to synthesise the encoded audio, and encoding side information, wherein side information comprises information indicating accessibility of side information, wherein the additional information is received as an audio parameter on the decoding processing side when the reliability of the predicted value of the calculated audio parameter is low.EFFECT: technical result consists in ensuring restoration of sound quality without increasing algorithmic delay, when packet loss occurs when encoding audio.3 cl, 43 dwg

Description

Technology area

[0001] The present invention relates to error concealment for transmitting audio packets over an IP or mobile network, and more specifically, relates to an audio encoding apparatus, an audio encoding method, an audio encoding program, an audio decoding apparatus, an audio decoding method, and an audio decoding program for high-precision generation of packet loss concealment signal to implement error concealment.

State of the art

[0002] When transmitting audio and acoustic signals (which are hereinafter collectively referred to as "audio signal") over an IP network or mobile network, the audio signal is encoded into audio packets at regular time intervals and transmitted over the communication network. On the receiving side, audio packets are received over a communication network and decoded into a decoded audio signal by a server, MCU (Multipoint Control Unit), terminal, or the like.

[0003] The audio signal is generally obtained in digital format. In particular, it is measured and collected as a sequence of digits, the number of which is the same as the sampling rate per second. Each element of the sequence is called a "sample". When encoding audio, each time a predetermined number of audio samples are collected in the embedded buffer, the audio signal is encoded in the buffer. The above specified number of samples is called "frame length" and a set with the same number of samples as the frame length is called "frame". For example, at a sampling rate of 32 kHz, when the frame length is 20 ms, the frame length is 640 samples. It should be noted that the buffer length can be more than one frame.

[0004] When transmitting audio packets over a communication network, a phenomenon (so-called "packet loss") may occur where some of the audio packets are lost, or an error may occur in some of the information recorded in the audio packets due to network congestion or the like. In such a case, the audio packets cannot be decoded correctly on the receiving side, and therefore, the desired decoded audio signal cannot be received. Additionally, the decoded audio signal corresponding to the audio packet where the packet loss occurred is detected as noise, which greatly degrades the subjective quality for the person listening to the audio.

[0005] In order to overcome the above-described disadvantages, packet loss concealment technique is used as a method for interpolating a portion of the audio / acoustic signal that is lost due to packet loss. There are two types of packet loss concealment technology: "packet loss concealment technology without the use of side information", where packet loss concealment is performed only on the receive side, and "packet loss concealment technology using side information", where parameters that help mask packet loss are obtained at the transmitting side and transmitted to the receiving side, where packet loss concealment is performed using the received parameters at the receiving side.

[0006] "Side information-free packet loss concealment technology" generates an audio signal corresponding to the portion where the packet loss occurred by copying the decoded audio signal contained in a packet that was correctly received in the past on a pitch by pitch basis. , and then multiplying it by some predetermined attenuation factor, for example, as described in Non-Patent Literature 1. Since the "packet loss concealment technique without using side information" is based on the assumption that the properties of the audio portion where the packet loss occurs are similar to those of the audio immediately before When the loss occurs, the masking effect cannot be sufficiently obtained when the portion of the audio where the packet loss occurred has properties other than audio just before the loss or when a sudden power change occurs.

[0007] On the other hand, "side information packet loss concealment technique" includes an approach that encodes parameters required for masking packet loss on the transmit side and transmits them for use in concealing packet loss on the receive side, as described in Patent Literature 1. In patent literature 1, audio is encoded by two encoding methods: main encoding and backup encoding. The spare coding encodes the frame immediately before the frame to be encoded with the base coding at a lower bit rate than the base coding (see FIG. 1 (a)). For example, the Nth packet contains an audio code obtained by coding an Nth frame with main coding and a side information code obtained by coding an (N-1) th frame with spare coding.

[0008] The receiving side waits for the arrival of two or more packets following one after the other in time and then decodes the earlier packet in time and receives the decoded audio signal. For example, to obtain a signal corresponding to the Nth frame, the receiving side waits for the (N + 1) th packet to arrive and then performs decoding. In the case where the Nth packet and the (N + 1) th packet are received correctly, the audio signal of the Nth frame is obtained by decoding the audio code contained in the Nth packet (see FIG. 1 (b)). On the other hand, in the case where a packet loss has occurred (when an (N + 1) th packet is received, provided that the N th packet is lost), the audio signal of the N th frame can be obtained by decoding the side information code contained in ( N + 1) th packet (see Fig. 1 (c)).

[0009] According to the method of Patent Literature 1, after the arrival of a packet to be decoded, to perform decoding, it is necessary to wait until one or more packets arrive and the algorithmic delay increases by one or more. Accordingly, in the method of Patent Literature 1, although the sound quality (audio quality) can be improved by masking the packet loss, the algorithmic delay increases, which causes the voice quality to deteriorate.

[0010] Additionally, in the case of applying the above-described packet loss concealment technique to CELP (Code Excited Linear Prediction) coding, another problem arises due to the characteristics of the CELP operation. Since CELP is a linear predictive audio model and is capable of high fidelity and high compression rate audio coding, it is used in many international standards.

[0011] In CELP, an audio signal is synthesized by filtering the excitation signal e (n) using a synthesis pole filter. In particular, the audio signal s (n) is synthesized according to the following equation:

Уравнение 1

Equation 1

where a (i) is the linear prediction coefficient (LP coefficient) and a value such as, for example, P = 16 is used as the degree.

[0012] The excitation signal is collected in a buffer called an adaptive codebook. When synthesizing audio for a new frame, an excitation signal is re-generated by adding an adaptive codebook vector read from the adaptive codebook and a fixed codebook vector representing the change in the excitation signal over time based on position information called pitch lag. The newly generated excitation signal is collected in an adaptive codebook and also filtered by a synthesis pole filter, and thereby a decoded signal is synthesized.

[0013] In CELP, the LP coefficient is calculated for all frames. When calculating the LP coefficient, a predictive signal of approximately 10 ms is required. Specifically, in addition to the frame to be encoded, the prediction signal is collected in a buffer, and then LP coefficient calculation and subsequent processing are performed (see FIG. 2). Each frame is divided into approximately four subframes, and processing such as the above-described pitch lag computation, adaptive codebook vector computation, fixed codebook vector computation, and adaptive codebook update are performed in each subframe. When processing each subframe, the LP coefficient is also interpolated so that the coefficient changes from subframe to subframe. Additionally, for quantization and interpolation, the LP coefficient is encoded after being converted into an ISP parameter (spectral immitance pair) and an ISF parameter (spectral imitation frequency), which are the equivalent representation (s) of the LP coefficient (s). A procedure for interconverting the LP coefficient (s) and the ISP parameter and the ISF parameter is described in non-patent literature 2.

[0014] In CELP encoding, encoding and decoding are performed on the assumption that both the encoding side and the decoding side have adaptive codebooks, and these adaptive codebooks are always synchronized with each other. Although the adaptive codebook on the coding side and the adaptive codebook on the decoding side are synchronized under the condition that packets are received correctly and decoding is performed correctly, once a packet loss has occurred, synchronization of the adaptive codebooks cannot be achieved.

[0015] For example, if the value that is used as the pitch lag is different between the encoding side and the decoding side, a time lag (time delay) occurs between the vectors of the adaptive codebook. Since the adaptive codebook is updated with these adaptive codebook vectors, even if the next frame is received correctly, the adaptive codebook vector computed on the encoding side and the adaptive codebook vector computed on the decoding side will not match, and the adaptive codebooks will not re-synchronize. Due to this incompatibility (inconsistency) of the adaptive codebooks, audio quality degrades for several frames after the frame where the packet loss occurred.

[0016] A more advanced technique for concealing packet loss in CELP coding is described in Patent Literature 2. According to Patent Literature 2, the transient codebook index is transmitted instead of pitch lag or adaptive codebook gain (gain) in some particular frame, which is largely degree damaged by packet loss. Patent Literature 2 technology focuses on a transition frame (transition from an audio segment without sound to an audio segment with sound, or a transition between two vowels) as a frame that is heavily damaged by packet loss. By generating an excitation signal using the transient codebook in this transient frame, it is possible to generate an excitation signal that is independent of the past adaptive codebook, and thereby recovering from the adaptive codebook incompatibility due to past packet loss.

[0017] However, since the method of Patent Literature 2 does not use a transition frame codebook in a frame in which, for example, a long vowel continues, it is not possible to recover from adaptive codebook incompatibilities in such a frame. Additionally, in the case where a packet containing a transition frame codebook is lost, the packet loss corrupts the frames after the loss. A similar situation occurs when the packet following the packet containing the transition frame codebook is lost.

[0018] Although it is possible to apply a codebook such as a transition frame codebook that does not depend on past frames to all frames, since the coding efficiency is significantly degraded, under these conditions it is not possible to achieve low bit rate and high sound quality. ...

List of links

Patent Literature

[0019] Patent Literature 1: Japanese Unexamined Patent Publication No. 2003-533916

Patent Literature 2: Japanese Unexamined Patent Publication No. 2010-507818

Non-patent literature

[0020] Non-patent literature 1: ITU-T G.711 Annex I

Non-patent Literature 2: 3GPP TS26-191

Non-patent Literature 3: 3GPP TS26-190

Non-patent Literature 4: ITU-T G.718

The essence of the invention

Technical problem

[0021] Using the method of Patent Literature 1, after the arrival of a packet to be decoded, decoding does not start until the next packet arrives. Thus, although the sound quality is improved by masking the packet loss, the algorithmic delay increases, which causes the degradation of the voice quality.

[0022] In the case of packet loss in CELP encoding, degradation in sound quality occurs due to the incompatibility of adaptive codebooks between the encoding unit and the decoding unit. Although the method described in Patent Literature 2 may allow recovery from adaptive codebook incompatibilities, the method is not sufficient to allow recovery when a frame other than the frame immediately before the transition frame is lost.

[0023] The present invention has been made to solve the above problems, and thus, an object of the present invention is to provide an audio coding apparatus, an audio coding method, an audio coding program, an audio decoding apparatus, an audio decoding method, and an audio decoding program that restore sound quality without increasing the algorithmic delays in case of packet loss when encoding audio.

Solution to the problem

[0024] To achieve the above object, according to one aspect of the present invention, there is provided an audio coding apparatus for coding an audio signal, which includes an audio coding unit configured to encode an audio signal and a side information (side) coding unit configured to calculate side information from predictive signal and coding of side information.

[0025] The side information may be associated with pitch lag in a predictive signal, associated with a pitch boost in a predictive signal, or associated with pitch lag and pitch gain in a predictive signal. Additionally, the side information may contain information related to the availability of the side information.

[0026] A side information coding unit may calculate side information for a part of a predicted signal and encode side information, and generate a masking signal, and the audio encoding apparatus may further include an error signal encoding unit configured to encode an error signal between an input audio signal and a masking signal outputted from the side information coding unit; and a main coding unit configured to encode the input audio signal.

[0027] Additionally, according to one aspect of the present invention, there is provided an audio decoding apparatus for decoding an audio code and outputting an audio signal, which includes an audio code buffer configured to detect packet loss based on an audio packet receive state, an audio parameter decoding unit configured to decode an audio code when the audio packet is received correctly, the side information decoding unit configured to decode the side information code when the audio packet is received correctly, the side information collecting unit configured to collect the side information obtained by decoding the side information code, the missing audio parameter processing unit configured to output an audio parameter when loss of an audio packet is detected, and an audio synthesis unit configured to synthesize decoded audio from the audio parameter.

[0028] The side information may be associated with pitch lag in a predictive signal, associated with a pitch boost in a predictive signal, or associated with pitch lag and pitch gain in a predictive signal. Additionally, the side information may contain information related to the availability of the side information.

[0029] The side information decoding unit may decode the side information code and output side information, and may further output a masking signal associated with the prediction portion by using the side information, and the audio decoding apparatus may further include an error decoding unit configured to decode of the code associated with the error signal between the audio signal and the masking signal, a main decoding unit configured to decode the code associated with the audio signal; and a masking signal collection unit configured to collect a masking signal output from the side information decoding unit.

[0030] When the audio packet is received correctly, a part of the decoded signal can be generated by adding a mask signal read from the mask signal acquisition unit and the decoded error signal output from the error decoding unit, and the mask signal acquisition unit can be updated with a mask signal output from side information decoding unit.

[0031] When the loss of an audio packet is detected, the masking signal read from the masking signal collector can be used as part or the whole of the decoded signal.

[0032] When the loss of an audio packet is detected, a decoded signal can be generated by using the audio parameter predicted by the missing audio parameter processing unit, and the masking signal collection unit can be updated using a part of the decoded signal.

[0033] When the loss of an audio packet is detected, the missing audio parameter processing unit may use the side information read from the side information collection unit as part of the predicted audio parameter value.

[0034] When the loss of an audio packet is detected, the audio synthesizer can correct the adaptive codebook vector, which is one of the audio parameters, by using side information read from the side information collector.

[0035] An audio coding method according to one aspect of the present invention is an audio coding method by an audio coding apparatus for coding an audio signal, which includes an audio coding step that encodes an audio signal and a side information encoding step that calculates side information from a prediction signal. and encode side information.

[0036] An audio decoding method according to one aspect of the present invention is an audio decoding method by an audio decoding apparatus for decoding an audio code and outputting an audio signal, which includes an audio code buffering step in which packet loss is detected based on an audio packet reception state, an audio parameter decoding step, on in which the audio code is decoded when the audio packet is received correctly, the side information decoding step, in which the side information code is decoded when the audio packet is received correctly, the side information gathering step, in which the side information obtained by decoding the side information code is collected, the missing audio parameter processing step, on which outputs an audio parameter when loss of an audio packet is detected; and an audio synthesis step that synthesizes decoded audio from the audio parameter.

[0037] An audio coding program according to one aspect of the present invention causes a computer to function as an audio coding unit for encoding an audio signal and a side information coding unit for calculating side information from a prediction signal and encoding side information.

[0038] An audio decoding program according to one aspect of the present invention causes a computer to function as an audio code buffer for detecting packet loss based on an audio packet reception state, an audio parameter decoding unit for decoding an audio code when the audio packet is received correctly, a side information decoding unit for decoding a side information code, when the audio packet is received correctly, a side information collecting unit for collecting side information obtained by decoding the side information code, a missing audio parameter processing unit for outputting an audio parameter when loss of an audio packet is detected, and an audio synthesis unit for synthesizing decoded audio from the audio parameter.

Benefits of the invention

[0039] Sound quality can be restored without increasing the algorithmic delay in the event of packet loss during audio encoding. Specifically, with CELP encoding, it is possible to reduce adaptive codebook degradation that occurs when a packet loss occurs, and thereby improve the sound quality in the event of a packet loss.

Brief Description of Drawings

[0040] FIG. 1 is a view showing the temporal relationship between packets and a decoded signal according to the related art described in Patent Literature 1.

FIG. 2 is a view showing a temporal relationship between an LP analysis target signal and a prediction signal when coded with CELP.

FIG. 3 is a view showing a temporal relationship between packets and a decoded signal according to an embodiment of the present invention.

FIG. 4 is a view showing an example of a functional configuration of an audio signal transmission apparatus in Example 1 (first example) of the present invention.

FIG. 5 is a view showing an example of a functional configuration of an audio signal receiving apparatus in Example 1 of the present invention.

FIG. 6 is a view showing a procedure of the audio signal transmission apparatus in Example 1 of the present invention.

FIG. 7 is a view showing a procedure of an audio signal receiving apparatus in Example 1 of the present invention.

FIG. 8 is a view showing an example of a functional configuration of a side information encoding unit in Example 1 of the present invention.

FIG. 9 is a view showing a procedure of a side information coding unit in Example 1 of the present invention.

FIG. 10 is a view showing the procedure of the LP coefficient calculating unit in Example 1 of the present invention.

FIG. 11 is a view showing a procedure of a target signal calculating unit in Example 1 of the present invention.

FIG. 12 is a view showing an example of a functional configuration of the audio missing parameter processing unit in Example 1 of the present invention.

FIG. 13 is a view showing an audio parameter prediction procedure in Example 1 of the present invention.

FIG. 14 is a view showing a procedure of the drive vector synthesis unit in Alternative Example 1-1 of Example 1 of the present invention.

FIG. 15 is a view showing an example of a functional configuration of an audio synthesis unit in Example 1 of the present invention.

FIG. 16 is a view showing the procedure of the audio synthesizer in Example 1 of the present invention.

FIG. 17 is a view showing an example of a functional configuration of a side information coding unit (when a side information output determining unit is included) in Alternative Example 1-2 of Example 1 of the present invention.

FIG. 18 is a view showing a procedure of a side information coding unit (when a side information output determining unit is included) in Alternative Example 1-2 of Example 1 of the present invention.

FIG. 19 is a view showing an audio parameter prediction procedure in Alternative Example 1-2 of Example 1 of the present invention.

FIG. 20 is a view showing an example of a functional configuration of an audio signal transmission apparatus in Example 2 of the present invention.

FIG. 21 is a view showing an example of a functional configuration of the main coding unit in Example 2 of the present invention.

FIG. 22 is a view showing a procedure of the audio signal transmission apparatus in Example 2 of the present invention.

FIG. 23 is a view showing an example of a functional configuration of an audio signal receiving apparatus in Example 2 of the present invention.

FIG. 24 is a view showing a procedure of an audio signal receiving apparatus in Example 2 of the present invention.

FIG. 25 is a view showing an example of a functional configuration of an audio synthesis unit in Example 2 of the present invention.

FIG. 26 is a view showing an example of a functional configuration of an audio parameter decoding unit in Example 2 of the present invention.

FIG. 27 is a view showing an example of a functional configuration of a side information encoding unit in Example 3 of the present invention.

FIG. 28 is a view showing a procedure of a side information coding unit in Example 3 of the present invention.

FIG. 29 is a view showing the procedure of the pitch lag selection unit in Example 3 of the present invention.

FIG. 30 is a view showing a procedure of a side information decoding unit in Example 3 of the present invention.

FIG. 31 is a view showing the configuration of an audio coding program and a storage medium according to an embodiment of the present invention.

FIG. 32 is a view showing the configuration of an audio decoding program and a storage medium according to an embodiment of the present invention.

FIG. 33 is a view showing an example of a functional configuration of a side information encoding unit in Example 4 of the present invention.

FIG. 34 is a view showing a procedure of a side information coding unit in Example 4 of the present invention.

FIG. 35 is a view showing a procedure of a pitch lag prediction block in Example 4 of the present invention.

FIG. 36 is another view showing the procedure of the pitch lag prediction unit in Example 4 of the present invention.

FIG. 37 is another view showing the procedure of the pitch lag prediction block in Example 4 of the present invention.

FIG. 38 is a view showing a procedure of an adaptive codebook calculator in Example 4 of the present invention.

FIG. 39 is a view showing an example of a functional configuration of a side information encoding unit in Example 5 of the present invention.

FIG. 40 is a view showing a procedure of a pitch lag coding unit in Example 5 of the present invention.

FIG. 41 is a view showing a procedure of a side information decoding unit in Example 5 of the present invention.

FIG. 42 is a view showing a procedure of a pitch lag prediction block in Example 5 of the present invention.

FIG. 43 is a view showing a procedure of an adaptive codebook calculator in Example 5 of the present invention.

Description of embodiments

[0041] Embodiments of the present invention are described hereinafter in this document with reference to the accompanying drawings. It should be noted that where possible, like elements are denoted with like reference numerals and redundant descriptions thereof are omitted.

[0042] An embodiment of the present invention relates to an encoder and a decoder that implement a "side information packet loss concealment technique" that encodes and transmits side information calculated on the encoder side for use in decoder side packet loss concealment.

[0043] In embodiments of the present invention, side information that is used to conceal a packet loss is contained in the previous (prior) packet. FIG. 3 shows the temporal relationship between the audio code and the side information code contained in the packet. As illustrated in FIG. 3, side information in embodiments of the present invention are parameters (pitch lag, adaptive codebook gain, and others) that are computed for a prediction signal when encoded with CELP.

[0044] Since the side information is contained in the previous packet, decoding can be performed without waiting for the packet that comes after the packet to be decoded. Additionally, when a packet loss is detected, since side information for a frame to be masked is obtained from a previous packet, highly accurate packet loss concealment can be implemented without waiting for the next packet.

[0045] In addition, by passing parameters for coding with CELP in the prediction signal as side information, it is possible to eliminate the adaptive codebook incompatibility even in case of packet loss.

[0046] Embodiments of the present invention may be composed of an audio signal transmitting apparatus (audio coding apparatus) and an audio signal receiving apparatus (audio decoding apparatus). An example of the functional configuration of the audio signal transmission apparatus is shown in FIG. 4 and an exemplary procedure for said device is shown in FIG. 6. Additionally, one example of a functional configuration of an audio signal receiving apparatus is shown in FIG. 5 and an exemplary procedure for said apparatus is shown in FIG. 7.

[0047] As shown in FIG. 4, the audio signal transmission apparatus includes an audio coding section 111 and a side information coding section 112. As shown in FIG. 5, the audio signal receiving apparatus includes an audio code buffer 121, an audio parameter decoding section 122, a missing audio parameter processing section 123, an audio synthesis section 124, a side information decoding section 125, and a side information collecting section 126.

[0048] An audio transmission apparatus encodes an audio signal for each frame, and can transmit an audio signal through the exemplary procedure shown in FIG. 6.

[0049] The audio coding unit 111 may calculate audio parameters for the frame to be encoded and output the audio code (step S131 in FIG. 6).

[0050] The side information encoding unit 112 may calculate audio parameters for the prediction signal and output the side information code (step S132 in FIG. 6).

[0051] It is determined whether the audio signal has ended, and the above steps may be repeated until the audio signal ends (step S133 in FIG. 6).

[0052] The audio signal receiving apparatus decodes the received audio packet and outputs the audio signal through the exemplary procedure shown in FIG. 7.

[0053] The audio code buffer 121 waits for an audio packet to arrive and collects the audio code. When the audio packet arrives correctly, processing switches to the audio parameter decoding unit 122. On the other hand, when the audio packet has not arrived correctly, the processing switches to the missing audio parameter processing unit 123 (step S141 in FIG. 7).

[0054] <When the audio packet is received correctly>

The audio parameter decoding unit 122 decodes the audio code and outputs the audio parameters (step S142 in FIG. 7).

[0055] The side information decoding unit 125 decodes the side information code and outputs the side information. The outputted side information is sent to the side information collecting unit 126 (step S143 in FIG. 7).

[0056] The audio synthesizing unit 124 synthesizes an audio signal from the audio parameters output from the audio parameter decoding unit 122, and outputs the synthesized audio signal (step S144 in FIG. 7).

[0057] The missing audio parameter processing unit 123 collects the audio parameters output from the audio parameter decoding unit 122 in case of a packet loss (step S145 in FIG. 7).

[0058] The audio code buffer 121 determines whether the transmission of audio packets is completed, and when the transmission of audio packets is completed, stops processing. While the transmission of audio packets continues, the above steps S141 to S146 are repeated (step S147 in FIG. 7).

[0059] <When audio packet is lost>

The audio missing parameter processing unit 123 reads side information from the side information collecting unit 126 and performs prediction on the parameter (s) not included in the side information, and thereby outputs the audio parameters (step S146 in FIG. 7).

[0060] The audio synthesis unit 124 synthesizes an audio signal from the audio parameters output from the missing audio parameter processing unit 123, and outputs the synthesized audio signal (step S144 in FIG. 7).

[0061] The missing audio parameter processing unit 123 collects the audio parameters output from the audio missing parameter processing unit 123 in case of packet loss (step S145 in FIG. 7).

[0062] The audio code buffer 121 determines whether the transmission of audio packets is complete, and when the transmission of audio packets is completed, stops processing. While the transmission of audio packets continues, the above steps S141 to S146 are repeated (step S147 in FIG. 7).

[0063] [Example 1]

In this example of a case in which the pitch lag is transmitted as side information, the pitch lag can be used to generate a packet loss concealment signal on the decoding side.

[0064] An example of a functional configuration of an audio signal transmission apparatus is shown in FIG. 4 and an example of the functional configuration of the audio signal receiving apparatus is shown in FIG. 5. An example of the procedure of the audio signal transmission apparatus is shown in FIG. 6 and an example of a procedure of an audio signal receiving apparatus is shown in FIG. 7.

[0065] <Transmission side>

In the audio signal transmission apparatus, the input audio signal is sent to the audio coding unit 111.

[0066] The audio coding unit 111 encodes a frame to be encoded by coding with CELP (step 131 in FIG. 6). For more details on CELP encoding, the method described in non-patent literature 3 is used as an example. Details of the CELP encoding procedure are omitted. It should be noted that in CELP encoding, local decoding is performed on the encoding side. Local decoding is for decoding audio code also on the coding side and obtaining parameters (ISP parameter and corresponding ISF parameter, pitch lag, long-term prediction parameter, adaptive codebook, adaptive codebook gain, fixed codebook gain, fixed codebook vector, and so on. ) required for audio synthesis. The parameters obtained by local decoding include at least one or both of an ISP parameter and an ISF parameter, a pitch lag and an adaptive codebook, which are sent to the side information encoding unit 112. In the case where the audio coding described in Non-Patent Literature 4 is used in the audio coding unit 111, an index representing characteristics of a frame to be encoded may also be sent to the side information coding unit 112. In embodiments, coding other than CELP coding may be used in audio coding unit 111. In embodiments using different coding, at least one or both of the ISP parameter and the ISF parameter, the pitch lag and the adaptive codebook may be separately computed from the input signal or the decoded local decoding signal and sent to the side coding unit 112. information.

[0067] The side information encoding unit 112 calculates the side information code using the parameters calculated by the audio encoding unit 111 and the prediction signal (step 132 in FIG. 6). As shown in the example of FIG. 8, the side information coding unit 112 includes an LP coefficient calculating unit 151, a target signal calculating unit 152, a pitch lag calculating unit 153, an adaptive codebook calculating unit 154, an excitation vector synthesis unit 155, an adaptive codebook buffer 156, a filter 157 synthesis and block 158 encoding the delay of the main tone. An exemplary procedure in a side information coding unit is shown in FIG. nine.

[0068] The LP coefficient calculating unit 151 calculates the LP coefficient using the ISF calculated by the audio coding unit 111 and the ISF calculated in the past several frames (step 161 in FIG. 9). The procedure of the LP coefficient calculating unit 151 is shown in FIG. ten.

в прогнозном сигнале. Параметр ISF

вычисляется следующим уравнением (этап 172 на Фиг. 10).[0069] First, the buffer is updated using the ISF parameter obtained from the audio coding unit 111 (step 171 in FIG. 10). Then, the ISF parameter is calculated

in the forecast signal. ISF parameter

is computed by the following equation (step 172 in FIG. 10).

Уравнение 2

Equation 2

Уравнение 3

Equation 3

является параметром ISF, сохраненным в буфере, который предназначен для кадра, которому предшествует j-число кадров. Дополнительно

является параметром ISF во время периода речи, который вычисляется заранее посредством обучения или подобным.

является константой и может быть значением, например, таким как 0,75, хотя не ограничивается им. Дополнительно

также является константой и может быть значением, например, таким как 0,9, хотя не ограничивается им.

,

и

могут меняться посредством индекса, представляющего характеристики кадра, который должен быть закодирован, как при маскировании ISF, описанном, например, в непатентной литературе 4.Where

is an ISF parameter stored in a buffer that is for a frame preceded by a j-number of frames. Additionally

is an ISF parameter during a speech period, which is calculated in advance by training or the like.

is constant and may be a value such as, for example, 0.75, but is not limited thereto. Additionally

is also a constant and can be a value such as, for example, 0.9, although not limited to it.

,

and

can be varied by an index representing the characteristics of the frame to be encoded, as in ISF concealment, described, for example, in non-patent literature 4.

удовлетворяет

, и значения

могут быть скорректированы так, чтобы смежное

не располагалось слишком близко. В качестве процедуры для корректировки значения

, может быть использована, например, непатентная литература 4 (уравнение 151) (этап 173 на Фиг. 10).[0070] In addition, the i values are arranged such that

satisfies

, and the values

can be adjusted so that the adjacent

was not too close. As a procedure for adjusting the value

, can be used, for example, non-patent literature 4 (equation 151) (step 173 in Fig. 10).

преобразуется в параметр ISP и интерполяция может быть выполнена для каждого подкадра. В качестве способа вычисления параметра ISP из параметра ISF может быть использован способ, описанный в разделе 6.4.4 в непатентной литературе 4, и в качестве способа интерполяции может быть использована процедура, описанная в разделе 6.8.3 в непатентной литературе 4 (этап 174 на Фиг. 10).[0071] After that, the ISF parameter

is mapped to an ISP parameter and interpolation can be performed for each subframe. As a method for calculating the ISP parameter from the ISF parameter, the method described in Section 6.4.4 in Non-Patent Literature 4 may be used, and as the interpolation method, the procedure described in Section 6.8.3 in Non-Patent Literature 4 may be used (step 174 in FIG. . ten).

. Число подкадров, содержащихся в прогнозном сигнале, равно

. Для преобразования из параметра ISP в коэффициент LP может быть использована процедура, описанная в разделе 6.4.5 в непатентной литературе 4 (этап 175 на Фиг. 10).[0072] Then, the ISP parameter for each subframe is converted to an LP factor

... The number of subframes contained in the predicted signal is

... The procedure described in Section 6.4.5 in Non-Patent Literature 4 (step 175 in FIG. 10) can be used to convert from ISP to LP.

(этап 162 на Фиг. 9). Как описано в разделе 6.8.4.1.3 в непатентной литературе 4, целевой сигнал получается посредством применения перцептивного взвешивающего фильтра к остаточному сигналу линейного предсказания (Фиг. 11).[0073] Target signal calculator 152 calculates target signal x (n) and impulse response h (n) by using coefficient LP

(step 162 in FIG. 9). As described in section 6.8.4.1.3 in Non-Patent Literature 4, the target signal is obtained by applying a perceptual weighting filter to the linear prediction residual signal (FIG. 11).

вычисляется с использованием коэффициента LP согласно следующему уравнению (этап 181 на Фиг. 11).[0074] First, the residual signal r (n) from the predictive signal

is calculated using the LP coefficient according to the following equation (step 181 in FIG. 11).

Уравнение 4

Equation 4

указывает число отсчетов подкадра, и L указывает число отсчетов кадра, который должен быть закодирован

. Затем удовлетворяется

.[0075] It should be noted that

indicates the number of samples of the subframe, and L indicates the number of samples of the frame to be encoded

... Then satisfied

...

вычисляется следующими уравнениями (этап 182 на Фиг. 11).[0076] In addition, the target signal

is computed by the following equations (step 182 in FIG. 11).

Уравнение 5

Equation 5

Уравнение 6

Equation 6

Уравнение 7

Equation 7

Уравнение 8

Equation 8

where the perceptual weighting filter ϒ = 0.68. The perceptual weighting filter value may be some other value according to the audio coding calculation technique.

вычисляется следующими уравнениями (этап 183 на Фиг. 11).[0077] Then the impulse response

is calculated by the following equations (step 183 in FIG. 11).

Уравнение 9

Equation 9

Уравнение 10

Equation 10

[0078] Pitch lag calculating unit 153 calculates the pitch lag for each subframe by calculating k that maximizes the following equation (step 163 in FIG. 9). It should be noted that in order to reduce the number of computations, the above-described target signal computation (step 182 in FIG. 11) and impulse response computation (step 183 in FIG. 11) may be omitted, and the residual signal may be used as the target signal. ...

Уравнение 11

Equation 11

Уравнение 12

Equation 12

Уравнение 13

Equation 13

получается посредством свертки импульсной характеристики с остатком линейного предсказания. Int(i) указывает интерполяционный фильтр. Детали интерполяционного фильтра являются такими, как описаны в разделе 6.8.4.1.4.1 в непатентной литературе 4. Как само собой разумеющееся

может быть применено без использования интерполяционного фильтра.It should be noted that

obtained by convolving the impulse response with a linear prediction residual. Int (i) indicates an interpolation filter. The details of the interpolation filter are as described in section 6.8.4.1.4.1 in the non-patent literature 4. As a matter of course

can be applied without using an interpolation filter.

[0079] Although the pitch lag can be calculated as an integer by the above-described calculation method, the pitch lag precision can be increased to the precision after the decimal point by interpolating the above T _k .

For the details of the procedure for calculating the pitch lag after the decimal point by interpolation, the processing method described in section 6.8.4.1.4.1 in the non-patent literature 4 may be used.

адаптивной кодовой книги и параметр долгосрочного предсказания из запаздывания T_p основного тона и адаптивной кодовой книги u(n), сохраненной в буфере 156 адаптивных кодовых книг, согласно следующему уравнению (этап 164 на Фиг. 9).[0080] The adaptive codebook calculator 154 computes a vector

adaptive codebook and long-term prediction parameter from pitch lag T _p and adaptive codebook u (n) stored in adaptive codebook buffer 156 according to the following equation (block 164 in FIG. 9).

Уравнение 14

Equation 14

For the details of the procedure for calculating the long-term parameter, the method described in section 5.7 in the non-patent literature 3 can be used.

адаптивной кодовой книги на некоторое предопределенное усиление

адаптивной кодовой книги и выводит вектор сигнала возбуждения согласно следующему уравнению (этап 165 на Фиг. 9).[0081] Block 155 synthesis of the excitation vector multiplies the vector

adaptive codebook for some predefined gain

adaptive codebook and outputs the excitation signal vector according to the following equation (step 165 in FIG. 9).

Уравнение 15

Equation 15

адаптивной кодовой книги может быть 1,0 или подобным, например, может быть использовано значение, заранее полученное обучением, или оно может быть изменено индексом, представляющим характеристики кадра, который должен быть закодирован.Although the gain value

adaptive codebook may be 1.0 or the like, for example, a value obtained in advance by training may be used, or it may be changed by an index representing characteristics of the frame to be encoded.

[0082] Next, the state of the adaptive codebook u (n) stored in the adaptive codebook buffer 156 is updated with the excitation vector according to the following equations (block 166 in FIG. 9).

Уравнение 16

Equation 16

Уравнение 17

Equation 17

[0083] The synthesis filter 157 synthesizes the decoded signal according to the following equation by inverse linear prediction filtering using the excitation signal vector as the excitation source (step 167 in FIG. 9).

Уравнение 18

Equation 18

[0084] The above steps 162 through 167 in FIG. 9 are repeated for each subframe until the end of the prediction signal (step 168 in FIG. 9).

основного тона, которое вычисляется в прогнозном сигнале (этап 169 на Фиг. 9). Число подкадров, содержащихся в прогнозном сигнале, равно

.[0085] The pitch lag encoding unit 158 encodes the pitch lag

the pitch that is calculated in the prediction signal (step 169 in FIG. 9). The number of subframes contained in the predicted signal is

...

[0086] The encoding may be performed in a manner such as, for example, one of the following techniques, although any technique may be used for encoding.

основного тона и передает результат.1. A method that performs binary coding, scalar quantization, vector quantization, or arithmetic coding on part or all of the delay

pitch and conveys the result.

из запаздывания основного тона предыдущего подкадра и передает результат, где

является запаздыванием основного тона последнего подкадра в кадре, который должен быть закодирован.2. A method that performs binary coding, scalar quantization, vector quantization, or arithmetic coding on part or all of the difference

from the pitch lag of the previous subframe and transmits the result, where

is the pitch lag of the last subframe in the frame to be encoded.

основного тона, и частью или всем запаздыванием основного тона, вычисленным для кадра, который должен быть закодирован, и передает результат.3. A method that performs vector quantization or arithmetic coding on either part or all of the delay

the pitch, and some or all of the pitch lag calculated for the frame to be encoded and conveys the result.

основного тона и передает индекс, указывающий выбранный способ интерполяции. В это время запаздывание основного тона множества подкадров, используемых для синтеза аудио в прошлом, также может быть использовано для выбора способа интерполяции.4. A method that selects one of a number of predefined interpolation methods based on part or all of the delay

pitch and transmits an index indicating the selected interpolation method. At this time, the pitch lag of the plurality of subframes used for audio synthesis in the past can also be used to select an interpolation method.

[0087] For scalar quantization and vector quantization, an empirically determined codebook or a codebook calculated in advance by training can be used. Additionally, a method that performs encoding after adding an offset value to the above pitch lag can also be included within the scope of an embodiment of the present invention as a matter of course.

[0088] <Decode side>

As shown in FIG. 5, an example of an audio signal receiving apparatus includes an audio code buffer 121, an audio parameter decoding section 122, a missing audio parameter processing section 123, an audio synthesis section 124, a side information decoding section 125, and a side information collecting section 126. The procedure of the audio signal receiving apparatus is as shown in the example of FIG. 7.

[0089] The audio code buffer 121 determines whether the packet is received correctly or not. When the audio code buffer 121 determines that the packet has been received correctly, processing switches to the audio parameter decoding section 122 and the side information decoding section 125. On the other hand, when the audio code buffer 121 determines that the packet has not been received correctly, the processing switches to the missing audio parameter processing unit 123 (step 141 in FIG. 7).

[0090] <When the packet is received correctly>

The audio parameter decoding unit 122 decodes the received audio code and calculates the audio parameters required to synthesize audio for the frame to be encoded (ISP parameter and corresponding ISF parameter, pitch lag, long-term prediction parameter, adaptive codebook, adaptive codebook gain, fixed codebook gain. books, a vector of a fixed codebook, and so on) (step 142 in FIG. 7).

основного тона и сохраняет его в блоке 126 сбора побочной информации. Блок 125 декодирования побочной информации декодирует код побочной информации посредством использования способа декодирования, соответствующего способу кодирования, используемому на стороне кодирования (этап 143 на Фиг. 7).[0091] The side information decoding unit 125 decodes the side information code, calculates the latency

the pitch and stores it in the block 126 collecting side information. The side information decoding unit 125 decodes the side information code by using a decoding method corresponding to the coding method used on the coding side (step 143 in FIG. 7).

[0092] The audio synthesizer 124 synthesizes an audio signal corresponding to a frame to be encoded based on parameters output from the audio parameter decoding unit 122 (step 144 in FIG. 7). An example of the functional configuration of the audio synthesis unit 124 is shown in FIG. 15 and an exemplary procedure of the audio synthesizer 124 is shown in FIG. 16. It should be noted that although the missing audio parameter processing unit 123 is illustrated showing a signal flow, the audio missing parameter processing unit 123 is not included in the functional configuration of the audio synthesis unit 124.

[0093] The LP coefficient calculating unit 1121 converts the ISF parameter into an ISP parameter and then performs interpolation processing and thereby obtains an ISP coefficient for each subframe. The LP coefficient calculator 1121 then converts the ISP coefficient to a linear prediction coefficient (LP coefficient), and thereby obtains the LP coefficient for each subframe (step 11301 in FIG. 16). To interpolate the ISP coefficient and the ISP-LP coefficient, the method described, for example, in section 6.4.5 in the non-patent literature 4 may be used. The parameter conversion procedure is not an essential part of an embodiment of the present invention and is thus not described in detail.

адаптивной кодовой книги вычисляется из запаздывания

основного тона и адаптивной кодовой книги u(n) согласно следующему уравнению.[0094] An adaptive codebook calculator 1123 calculates an adaptive codebook vector by using a pitch lag, a long-term prediction parameter, and an adaptive codebook 1122 (block 11302 in FIG. 16). Vector

adaptive codebook computed from lag

pitch and adaptive codebook u (n) according to the following equation.

Уравнение 19

Equation 19

. Фильтр Int(i), который используется для интерполирования, является таким же как интерполяционный фильтр из The adaptive codebook vector is computed by interpolating the adaptive codebook u (n) using the FIR Int (i) filter. The length of the adaptive codebook is

... The Int (i) filter that is used for interpolation is the same as the interpolation filter from

Уравнение 20

Equation 20

является числом отсчетов подкадра. Нет необходимости использовать фильтр для интерполирования, несмотря на то, что на стороне кодера фильтр используется для интерполирования.It is an FIR filter with some predefined length 2l + 1.

is the number of samples of the subframe. It is not necessary to use a filter for interpolation, although on the encoder side a filter is used for interpolation.

[0095] The adaptive codebook calculator 1123 filters the adaptive codebook vector according to the long-term prediction parameter value (block 11303 in FIG. 16). When the long-term prediction parameter has a value indicating that filtering is enabled, filtering is performed on the adaptive codebook vector by the following equation.

v` (n) = 0.18v` (n-1) + 0.64v` (n) + 0.18v` (n + 1) Equation 21

.[0096] On the other hand, when the long term prediction parameter has a value indicating that there is no need for filtering, filtering is not performed and set

...

адаптивной кодовой книги (этап 11304 на Фиг. 16). Дополнительно, блок 1124 синтеза вектора возбуждения умножает вектор c(n) фиксированной кодовой книги на усиление

фиксированной кодовой книги (этап 11305 на Фиг. 16). Кроме того, блок 1124 синтеза вектора возбуждения складывает вместе вектор адаптивной кодовой книги и вектор фиксированной кодовой книги и выводит вектор сигнала возбуждения (этап 11306 на Фиг. 16).[0097] Excitation vector synthesis unit 1124 multiplies the adaptive codebook vector by the gain

adaptive codebook (block 11304 in FIG. 16). Additionally, block 1124 synthesis of the excitation vector multiplies the vector c (n) of the fixed codebook by the gain

a fixed codebook (block 11305 in FIG. 16). In addition, the excitation vector synthesizer 1124 adds the adaptive codebook vector and the fixed codebook vector together and outputs the excitation signal vector (step 11306 in FIG. 16).

Уравнение 22

Equation 22

[0098] Post-filter 1125 performs post-processing such as pitch enhancement, noise cancellation, and low-frequency enhancement, for example, on the excitation signal vector. Details of methods such as pitch improvement, noise elimination and low frequency enhancement are described in Section 6.1 in Non-Patent Literature 3. Post-filter processing is not substantially related to an essential part of an embodiment of the present invention, and thus is not described in detail (step 11307 in FIG. . 16).

[0099] Adaptive codebook 1122 updates the state with the excitation signal vector according to the following equations (block 11308 in FIG. 16).

Уравнение 23

Equation 23

Уравнение 24

Equation 24

[0100] The synthesis filter 1126 synthesizes the decoded signal according to the following equation by inverse linear prediction filtering using the excitation signal vector as the excitation source (step 11309 in FIG. 16).

Уравнение 25

Equation 25

[0101] A perceptual inverse weighting filter 1127 applies a perceptual inverse weighting filter to the decoded signal according to the following equation (step 11310 in FIG. 16).

Уравнение 26

Equation 26

типично равно 0,68 или подобному, хотя не ограничивается этим значением.Value

typically 0.68 or the like, although not limited to this value.

[0102] Missing audio parameters processing unit 123 stores audio parameters (ISF parameter, pitch lag, adaptive codebook gain, fixed codebook gain) used in buffer audio synthesizer 124 (block 145 in FIG. 7).

[0103] <When packet loss is detected>

основного тона из блока 126 сбора побочной информации и предсказывает аудиопараметры. Пример функциональной конфигурации блока 123 обработки отсутствующих аудиопараметров показан в примере с Фиг. 12 и примерная процедура предсказания аудиопараметров показана на Фиг. 13.Missing audio parameters processing unit 123 reads the delay

the pitch from the side information collector 126 and predicts audio parameters. An example of the functional configuration of the missing audio parameter processing unit 123 is shown in the example of FIG. 12 and an exemplary audio parameter prediction procedure is shown in FIG. 13.

[0104] ISF prediction unit 191 calculates the ISF parameter using the ISF parameter for the previous frame and the ISF parameter calculated for the past several frames (block 1101 in FIG. 13). The procedure of the ISF prediction block 191 is shown in FIG. ten.

вычисляется согласно следующему уравнению (этап 172 на Фиг. 10).[0105] First, the buffer is updated using the ISF parameter from the immediately preceding frame (step 171 in FIG. 10). Further parameter ISF

is calculated according to the following equation (step 172 in FIG. 10).

Уравнение 27

Equation 27

Уравнение 28

Equation 28

является параметром ISF, сохраненным в буфере, который предназначен для кадра, которому предшествует j-число кадров. Дополнительно

,

и

являются такими же значениями, как значения, используемые на стороне кодирования.Where

is an ISF parameter stored in a buffer that is for a frame preceded by a j-number of frames. Additionally

,

and

are the same values as the values used on the encoding side.

удовлетворяет

и значения

скорректированы так, чтобы смежное

не располагалось слишком близко. В качестве примерной процедуры для корректировки значения

может быть использована непатентная литература 4 (уравнение 151) (этап 173 на Фиг. 10).[0106] In addition, the values of i are arranged such that

satisfies

and values

adjusted so that the adjacent

was not too close. As an example procedure for adjusting the value

non-patent literature 4 (equation 151) may be used (step 173 in FIG. 10).

основного тона. Дополнительно посредством использования запаздывания

основного тона, использованного для прошлого декодирования, блок 192 предсказания запаздывания основного тона выводит запаздывание

основного тона. Число подкадров, содержащихся в одном кадре, равно M и число запаздываний основного тона, содержащихся в побочной информации, равно

. Для предсказания запаздывания

основного тона может быть использована процедура, описанная, например, в разделе 7.11.1.3 в непатентной литературе 4 (этап 1102 на Фиг. 13).[0107] The pitch lag prediction unit 192 decodes the side information code from the side information collection unit 126, and thereby obtains the latency

the main tone. Additionally through the use of lag

the pitch used for past decoding, the pitch lag prediction unit 192 outputs a lag

the main tone. The number of subframes contained in one frame is M and the number of pitch lags contained in the side information is

... To predict lag

the pitch may be followed by the procedure described, for example, in section 7.11.1.3 in non-patent literature 4 (step 1102 in Fig. 13).

адаптивной кодовой книги посредством использования некоторого предопределенного усиления

адаптивной кодовой книги и усиления

адаптивной кодовой книги, использованного при прошлом декодировании. Число подкадров, содержащихся в одном кадре, равно M и число запаздываний основного тона, содержащихся в побочной информации, равно

. Для предсказания усиления

адаптивной кодовой книги может быть использована процедура, описанная, например, в разделе 7.11.2.5.3 в непатентной литературе 4 (этап 1103 на Фиг. 13).[0108] Adaptive codebook gain predictor 193 outputs the gain

adaptive codebook by using some predefined gain

adaptive codebook and gain

adaptive codebook used in past decoding. The number of subframes contained in one frame is M and the number of pitch lags contained in the side information is

... To predict gain

adaptive codebook, the procedure described, for example, in section 7.11.2.5.3 in Non-Patent Literature 4 (block 1103 in FIG. 13) may be used.

фиксированной кодовой книги посредством использования усиления

фиксированной кодовой книги, использованного при прошлом декодировании. Число подкадров, содержащихся в одном кадре, равно M. Для предсказания усиления

фиксированной кодовой книги может быть использована процедура, описанная в разделе 7.11.2.6 в непатентной литературе 4, например (этап 1104 на Фиг. 13).[0109] Fixed codebook gain predictor 194 outputs the gain

fixed codebook by using gain

the fixed codebook used in past decoding. The number of subframes contained in one frame is M. For gain prediction

fixed codebook, the procedure described in section 7.11.2.6 in Non-Patent Literature 4 may be used, for example (step 1104 in FIG. 13).

[0110] Block 195 generating noise signals outputs a vector of noise, such as white noise, with a length L (step 1105 in Fig. 13). The length of one frame is L.

[0111] The audio synthesis unit 124 synthesizes the decoded signal based on the audio parameters output from the missing audio parameter processing unit 123 (step 144 in FIG. 7). The operation of the audio synthesizer 124 is the same as the operation of the audio synthesizer <When audio packet is received correctly> and is not described in detail again (step 144 in FIG. 7).

[0112] Missing audio parameters processing unit 123 stores audio parameters (ISF parameter, pitch lag, adaptive codebook gain, fixed codebook gain) used in audio synthesizer 124 to buffer (block 145 in FIG. 7).

[0113] Although the case of encoding and transmitting side information for all subframes contained in the prediction signal is described in the above example, a configuration that transmits only side information for some specific subframe can be used.

[0114] [Alternative example 1-1]

As an alternative example to the previously discussed Example 1, an example is described hereinafter in this document that adds pitch enhancement to side information. The difference between Alternative Example 1-1 and Example 1 is only the operation of the drive vector synthesis unit 155, and therefore the description of other parts is omitted.

[0115] <Coding side>

The procedure of the excitation vector synthesis unit 155 is shown in the example of FIG. 14.

адаптивной кодовой книги вычисляется из вектора

адаптивной кодовой книги и целевого сигнала x(n), согласно следующему уравнению (этап 1111 на Фиг. 14).[0116] Gain

adaptive codebook is computed from the vector

adaptive codebook and target signal x (n) according to the following equation (block 1111 in FIG. 14).

ограниченному посредством 0≤g_p≤1,2 Уравнение 29

limited by 0≤g _p ≤1.2 Equation 29

является сигналом

, который получается посредством свертки импульсной характеристики с вектором адаптивной кодовой книги.Where

is a signal

which is obtained by convolving the impulse response with an adaptive codebook vector.

[0117] The calculated adaptive codebook gain is encoded and contained in the side information code (block 1112 in FIG. 14). For encoding, scalar quantization can be used using a codebook obtained in advance through training, although any other technique can also be used for encoding.

адаптивной кодовой книги, полученное посредством декодирования кода, вычисленного при кодировании усиления адаптивной кодовой книги, вектор возбуждения вычисляется согласно следующему уравнению (этап 1113 на Фиг. 14).[0118] By multiplying the adaptive codebook vector by the gain

of the adaptive codebook obtained by decoding the code calculated by encoding the gain of the adaptive codebook, the excitation vector is calculated according to the following equation (step 1113 in FIG. 14).

Уравнение 30

Equation 30

[0119] <Decode side>

адаптивной кодовой книги на усиление

адаптивной кодовой книги, полученное посредством декодирования кода побочной информации, и выводит вектор сигнала возбуждения согласно следующему уравнению (этап 165 на Фиг. 9).Block 155 synthesis of the excitation vector multiplies the vector

adaptive gain codebook

adaptive codebook obtained by decoding the side information code, and outputs the excitation signal vector according to the following equation (step 165 in FIG. 9).

Уравнение 31

Equation 31

[0120] [Alternative example 1-2]

As an alternative example for example 1, hereinafter in this document, an example is described that adds a flag to determine the use of side information to side information.

[0121] <Coding side>

An example of the functional configuration of the side information coding unit is shown in FIG. 17 and the procedure of the side information coding unit is shown in the example of FIG. 18. The only difference from example 1 is the side information output determination unit 1128 (step 1131 in FIG. 18) and therefore the description of other parts is omitted.

[0122] Side information output determining unit 1128 calculates the segment signal-to-noise ratio (SNR) of the decoded signal and the predicted signal according to the following equation, and only when the segment SNR exceeds a certain threshold, sets the flag value to "ON" and adds it to the side information.

Уравнение 32

Equation 32

On the other hand, when the segment SNR does not exceed the threshold, the side information output determination unit 1128 sets the flag value to "OFF" and adds it to the side information (step 1131 in FIG. 18). It should be noted that the number of side information bits can be reduced by adding side information such as pitch lag and pitch gain to the flag and transmitting the added side information only when the flag value is "ON", and transmitting only the flag value when the value flag - "OFF".

[0123] <Decode side>

The side information decoding unit decodes a flag contained in the side information code. When the flag value is "ON", the audio missing parameter processing unit calculates the decoded signal using the same procedure as in Example 1. On the other hand, when the flag value is "OFF", it calculates the decoded signal using packet loss concealment technique without using side information (block 1151 in Fig. 19).

[0124] [Example 2]

подкадра (подкадров).This example also uses the decoded audio of the prediction portion when the packet is correctly received. For the purposes of this discussion, the number of subframes contained in one frame is M subframes and the prediction signal length is

subframes (subframes).

[0125] <Coding side>

As shown in the example of FIG. 20, the audio signal transmission apparatus includes a main coding unit 211, a side information coding unit 212, a masking signal collection unit 213, and an error signal coding unit 214. The procedure of the audio signal transmission apparatus is shown in FIG. 22.

[0126] The error signal coding unit 214 reads the masking signal for one subframe from the masking signal acquisition unit 213, subtracts it from the audio signal, and thereby calculates the error signal (step 221 in FIG. 22).

[0127] The error signal coding unit 214 encodes the error signal. As some specific example procedure, the AVQ described in section 6.8.4.1.5 in Non-Patent Literature 4 may be used. When the error signal is encoded, local decoding is performed and the decoded error signal is output (step 222 in FIG. 22).

[0128] By adding the decoded error signal to the masking signal, a decoded signal for one subframe is outputted (step 223 in FIG. 22).

подкадров до конца сигнал маскирования.[0129] The above steps 221 to 223 are repeated for

subframes until the end of the masking signal.

[0130] An exemplary functional configuration of the main coding unit 211 is shown in FIG. 21. The main coding unit 211 includes an ISF coding unit 2011, a target signal calculating unit 2012, a pitch lag calculating unit 2013, an adaptive codebook calculating unit 2014, a fixed codebook calculating unit 2015, a gain calculating unit 2016, a vector calculating unit 2017 excitation, synthesis filter 2018 and adaptive codebook buffer 2019.

[0131] The ISF coding unit 2011 obtains the LP coefficient by applying the Levinson-Durbin method to the frame to be encoded and the prediction signal. The ISF encoder 2011 then converts the LP coefficient into an ISF parameter and encodes the ISF parameter. The ISF encoder 2011 then decodes the code and obtains the decoded ISF parameter. Ultimately, the ISF coding unit 2011 interpolates the decoded ISF parameter and obtains the decoded LP coefficient for each subframe. The procedures of the Levinson-Durbin method and the conversion from LP coefficient to ISF parameter are the same as in Example 1. Additionally, the procedure described, for example, in section 6.8.2 in the non-patent literature 4, can be used to encode the ISF parameter. by encoding the ISF parameter, the decoded ISF parameter, and the decoded LP coefficient (which is obtained by converting the decoded ISF parameter to the LP coefficient) can be obtained by the ISF encoding unit 2011 (step 224 in FIG. 22).

[0132] The detailed procedure of the target signal calculator 2012 is the same as in step 162 in FIG. 9 in Example 1 (step 225 in FIG. 22).

[0133] The pitch lag calculating unit 2013 refers to the adaptive codebook buffer, and calculates the pitch lag and the long-term prediction parameter by using the target signal. The detailed procedure for calculating the pitch lag and long term prediction parameter is the same as in Example 1 (step 226 in FIG. 22).

[0134] The adaptive codebook calculator 2014 calculates the adaptive codebook vector by using the pitch lag and the long-term prediction parameter calculated by the pitch lag calculating section 2013. The detailed procedure of the adaptive codebook calculator 2014 is the same as in Example 1 (step 227 in FIG. 22).

[0135] The fixed codebook calculator 2015 calculates the fixed codebook vector and the index obtained by encoding the fixed codebook vector by using the target signal and the adaptive codebook vector. The detailed procedure is the same as the AVQ procedure used in the error signal coding unit 214 (step 228 in FIG. 22).

[0136] Gain calculator 2016 calculates an adaptive codebook gain, a fixed codebook gain, and an index obtained by encoding the two gains using a target signal, an adaptive codebook vector, and a fixed codebook vector. A detailed procedure that can be used is described, for example, in section 6.8.4.1.6 in Non-Patent Literature 4 (step 229 in FIG. 22).

[0137] The excitation vector calculator 2017 calculates the excitation vector by adding the adaptive codebook vector and the fixed codebook vector to which the gain is applied. The detailed procedure is the same as in Example 1. Then, the excitation vector calculator 2017 updates the state of the adaptive codebook buffer 2019 by using the excitation vector. The detailed procedure is the same as in example 1 (step 2210 in FIG. 22).

[0138] The synthesis filter 2018 synthesizes the decoded signal by using the decoded LP coefficient and the excitation vector (block 2211 in FIG. 22).

до конца кадра, который должен быть закодирован.[0139] The above steps 224 to 2211 are repeated for subframes

to the end of the frame to be encoded.

прогнозного сигнала. Конкретная процедура является такой же, как в примере 1 (этап 2212 на Фиг. 22).[0140] Side information coding unit 212 calculates side information for a subframe

predictive signal. The specific procedure is the same as in example 1 (step 2212 in FIG. 22).

[0141] In addition to the procedure of Example 1, the decoded signal outputted by the synthesis filter 157 of the side information encoding unit 212 is collected in the masking signal acquisition unit 213 in Example 2 (step 2213 in FIG. 22).

[0142] <Decoding unit>

As shown in FIG. 23, an example of an audio signal receiving apparatus includes an audio code buffer 231, an audio parameter decoding unit 232, a missing audio parameter processing unit 233, an audio synthesis unit 234, a side information decoding unit 235, a side information collection unit 236, an error signal decoding unit 237, and an acquisition unit 238 masking signals. An exemplary procedure for an audio signal receiving apparatus is shown in FIG. 24. An exemplary functional configuration of the audio synthesis unit 234 is shown in FIG. 25.

[0143] The audio code buffer 231 determines whether the packet is received correctly or not. When the audio code buffer 231 determines that the packet has been received correctly, the processing switches to the audio parameter decoding section 232, the side information decoding section 235, and the error signal decoding section 237. On the other hand, when the audio code buffer 231 determines that a packet has not been received correctly, processing switches to the missing audio parameter processing unit 233 (step 241 in FIG. 24).

[0144] <When the packet is received correctly>

The error signal decoding unit 237 decodes the error signal code and obtains the decoded error signal. As a specific example procedure, a decoding method corresponding to a method used on the encoding side such as AVQ described in section 7.1.2.1.2 in Non-Patent Literature 4 (step 242 in FIG. 24) may be used.

[0145] The excitation predictive vector synthesis unit 2318 reads a masking signal for one subframe from the masking signal acquisition unit 238 and adds a masking signal to the decoded error signal and thereby outputs the decoded signal for one subframe (step 243 in FIG. 24).

до окончания сигнала маскирования.[0146] The above steps 241 to 243 are repeated for subframes

until the end of the masking signal.

[0147] The audio parameter decoding unit 232 includes an ISF decoding unit 2211, a pitch lag decoding unit 2212, a gain decoding unit 2213, and a fixed codebook decoding unit 2214. An example of the functional configuration of the audio parameter decoding unit 232 is shown in FIG. 26.

[0148] The ISF decoding unit 2211 decodes the ISF code and converts it to the LP coefficient, and thereby obtains the decoded LP coefficient. For example, the procedure described in section 7.1.1 in Non-Patent Literature 4 (step 244 in FIG. 24) is used.

[0149] The pitch lag decoding unit 2212 decodes the pitch lag code and obtains the pitch lag and the long-term prediction parameter (block 245 in FIG. 24).

[0150] The gain decoding unit 2213 decodes the gain code and obtains the adaptive codebook gain and the fixed codebook gain. An exemplary detailed procedure is described in section 7.1.2.1.3 in Non-Patent Literature 4 (step 246 in FIG. 24).

[0151] An adaptive codebook calculator 2313 calculates an adaptive codebook vector by using a pitch lag and a long-term prediction parameter. The detailed procedure of the adaptive codebook calculator 2313 is described in Example 1 (step 247 in FIG. 24).

[0152] The fixed codebook decoding unit 2214 decodes the fixed codebook code and calculates the fixed codebook vector. The detailed procedure is described in section 7.1.2.1.2 in Non-Patent Literature 4 (step 248 in FIG. 24).

[0153] An excitation vector synthesis unit 2314 calculates an excitation vector by adding the adaptive codebook vector and the fixed codebook vector to which the gain is applied. Additionally, the excitation vector calculator updates the adaptive codebook buffer by using the excitation vector (step 249 in FIG. 24). The detailed procedure is the same as in example 1.

[0154] The synthesis filter 2316 synthesizes the decoded signal by using the decoded LP coefficient and the excitation vector (block 2410 in FIG. 24). The detailed procedure is the same as in example 1.

до конца кадра, который должен быть закодирован.[0155] The above steps 244 to 2410 are repeated for subframes

to the end of the frame to be encoded.

[0156] The functional configuration of the side information decoding section 235 is the same as in Example 1. The side information decoding section 235 decodes the side information code and calculates the pitch lag (step 2411 in FIG. 24).

[0157] The functional configuration of the missing audio parameter processing unit 233 is the same as in Example 1.

The ISF prediction unit 191 predicts the ISF parameter using the ISF parameter for the previous frame and converts the predicted ISF parameter to the LP coefficient. The procedure is the same as in steps 172, 173, and 174 of Example 1 shown in FIG. 10 (block 2412 in FIG. 24).

[0158] The adaptive codebook calculator 2313 calculates the adaptive codebook vector by using the pitch lag output from the side information decoding unit 235 and the adaptive codebook 2312 (step 2413 in FIG. 24). The procedure is the same as in steps 11301 and 11302 in FIG. 16.

[0159] An adaptive codebook gain predictor 193 outputs the adaptive codebook gain. The specific procedure is the same as step 1103 in FIG. 13 (block 2414 in FIG. 24).

[0160] Fixed codebook gain predictor 194 outputs the fixed codebook gain. The specific procedure is the same as step 1104 in FIG. 13 (block 2415 in FIG. 24).

[0161] The noise signal generating unit 195 outputs noise such as white noise as a fixed codebook vector. The procedure is the same as at 1105 in FIG. 13 (block 2416 in FIG. 24).

[0162] The excitation vector synthesizer 2314 applies a gain to each of the adaptive codebook vector and the fixed codebook vector and adds them together and thereby calculates the excitation vector. Additionally, excitation vector synthesizer 2314 updates the adaptive codebook buffer with the excitation vector (block 2417 in FIG. 24).

[0163] The synthesis filter 2316 calculates the decoded signal using the above-described LP coefficient and the excitation vector. The synthesis filter 2316 then updates the masking signal acquisition block 238 with the calculated decoded signal (block 2418 in FIG. 24).

и декодированный сигнал выводится в качестве аудиосигнала.[0164] The above steps are repeated for subframes

and the decoded signal is output as an audio signal.

[0165] <When packet is lost>

A masking signal for one subframe is read from the masking signal collector and used as a decoded signal (step 2419 in FIG. 24).

.[0166] The above is repeated for subframes

...

[0167] The ISF prediction unit 191 predicts the ISF parameter (block 2420 in FIG. 24). As a routine, block 1101 in FIG. 13.

[0168] Pitch lag predictor 192 outputs the predicted pitch lag by using the pitch lag used in past decoding (step 2421 in FIG. 24). The procedure used for prediction is the same as at 1102 in FIG. 13.

[0169] The operations of the adaptive codebook gain predictor 193, the fixed codebook gain predictor 194, the noise generating unit 195, and the audio synthesis unit 234 are the same as in Example 1 (block 2422 in FIG. 24).

выводится в качестве аудиосигнала, и блок 238 сбора сигналов маскирования обновляется декодированным сигналом для оставшихся подкадров

.[0170] The above steps are repeated for the M subframes and the decoded signal for the subframes

is output as an audio signal, and the masking signal acquisition unit 238 is updated with the decoded signal for the remaining subframes

...

[0171] [Example 3]

Hereinafter, this document describes a case using guttural pulse synchronization when calculating the adaptive codebook vector.

[0172] <Coding side>

The functional configuration of the audio signal transmission apparatus is the same as in Example 1. The functional configuration and procedure differ only in the side information coding unit, and therefore, only the operation of the side information coding unit will be described below.

[0173] The side information coding unit includes an LP coefficient calculating unit 311, a pitch lag prediction unit 312, a pitch lag selection unit 313, a pitch lag coding unit 314, and an adaptive codebook buffer 315. An example of the functional configuration of the side information coding unit is shown in FIG. 27 and an exemplary procedure of the side information coding unit is shown in the example of FIG. 28.

[0174] The LP coefficient calculating unit 311 is the same as the LP coefficient calculating unit in Example 1, and thus will not be described again (step 321 in FIG. 28).

запаздывания основного тона с использованием запаздывания основного тона, полученного из блока кодирования аудио (этап 322 на Фиг. 28). Конкретная обработка предсказания является такой же, как предсказание запаздывания

основного тона в блоке 192 предсказания запаздывания основного тона в примере 1 (которое является таким же, как на этапе 1102 на Фиг. 13).[0175] Pitch lag prediction unit 312 calculates a predicted value

a pitch lag using the pitch lag obtained from the audio coding unit (step 322 in FIG. 28). The specific prediction processing is the same as lag prediction

the pitch in the pitch lag prediction block 192 in Example 1 (which is the same as in step 1102 in FIG. 13).

[0176] Next, the pitch lag selection unit 313 determines the pitch lag to be transmitted as side information (step 323 in FIG. 28). The detailed procedure of the pitch lag selection block 313 is shown in the example of FIG. 29.

запаздывания основного тона и значения прошлого запаздывания

основного тона согласно следующим уравнениям (этап 331 на Фиг. 29).[0177] First, a pitch lag codebook is generated from the predicted value

pitch lags and past lag values

the pitch according to the following equations (step 331 in FIG. 29).

><When

>

Уравнение 33

Equation 33

><When

>

Уравнение 34

Equation 34

. Дополнительно, число индексов кодовой книги равно I.

является некоторой предопределенной дискретой и

является некоторой предопределенной константой.The pitch lag value one subframe earlier is

... Additionally, the number of codebook indices is I.

is some predefined discrete and

is some predefined constant.

запаздывания основного тона, исходный вектор

возбуждения генерируется согласно следующему уравнению (этап 332 на Фиг. 29).[0178] Then, by using the adaptive codebook and the predicted value

pitch lag, original vector

excitation is generated according to the following equation (step 332 in FIG. 29).

Уравнение 35

Equation 35

The procedure for calculating the original excitation vector is the same as equations (607) and (608) in non-patent literature 4.

основного тона в кодовой книги запаздываний основного тона, чтобы посредством этого сгенерировать потенциальный вектор

адаптивной кодовой книги (этап 333 на Фиг. 29). Для гортанной импульсной синхронизации может быть использована такая же процедура, как в случае, описанном в разделе 7.11.2.5 в непатентной литературе 4, где положение импульса не является доступным. Следует отметить, однако, что u(n) в непатентной литературе 4 соответствует

в варианте осуществления настоящего изобретения, и экстраполированный основной тон соответствует

в варианте осуществления настоящего изобретения, и последний достоверный основной тон (

) соответствует

в варианте осуществления настоящего изобретения.[0179] Then, laryngeal pulse synchronization is applied to the original excitation vector by using all potential delays

pitch in the pitch lag codebook to thereby generate a potential vector

adaptive codebook (block 333 in FIG. 29). For laryngeal pulse synchronization, the same procedure can be used as in the case described in section 7.11.2.5 in the non-patent literature 4, where the position of the pulse is not available. It should be noted, however, that u (n) in the non-patent literature 4 corresponds to

in an embodiment of the present invention, and the extrapolated pitch corresponds to

in an embodiment of the present invention, and the last valid pitch (

) matches

in an embodiment of the present invention.

адаптивной кодовой книги вычисляется шкала оценок (этап 334 на Фиг. 29). В случае использования сегментного SNR в качестве шкалы оценок, сигнал синтезируется посредством обратной фильтрации с использованием коэффициента LP и сегментное SNR вычисляется с входным сигналом согласно следующему уравнению.[0180] For a potential vector

the adaptive codebook calculates the rating scale (block 334 in FIG. 29). In the case of using the segment SNR as the rating scale, the signal is synthesized by inverse filtering using the LP coefficient, and the segment SNR is calculated with the input signal according to the following equation.

Уравнение 35

Equation 35

Уравнение 36

Equation 36

[0181] Instead of performing inverse filtering, the segment SNR can be calculated in the adaptive codebook vector domain by using the residual signal according to the following equation.

Уравнение 37

Equation 37

Уравнение 38

Equation 38

вычисляется посредством использования коэффициента LP (этап 181 на Фиг. 11).In this case, the residual signal r (n) from the predictive signal

computed by using the LP coefficient (step 181 in FIG. 11).

[0182] The index corresponding to the highest rating scale computed in step 334 is selected, and the pitch lag corresponding to the index is calculated (step 335 in FIG. 29).

Уравнение 39

Equation 39

[0183] <Decode side>

The functional configuration of the audio signal receiving apparatus is the same as in Example 1. Differences from Example 1 are the functional configuration and procedure of the missing audio parameter processing unit 123, the side information decoding unit 125 and the side information collecting unit 126, and only these are described later in this document. text.

[0184] <When packet is received correctly>

основного тона и сохраняет его в блок 126 сбора побочной информации. Примерная процедура блока 125 декодирования побочной информации показана на Фиг. 30.The side information decoding unit 125 decodes the side information code and calculates the delay

pitch and stores it in the block 126 collecting side information. An exemplary procedure for side information decoding unit 125 is shown in FIG. thirty.

запаздывания основного тона посредством использования запаздывания основного тона, полученного из блока декодирования аудио (этап 341 на Фиг. 30). Конкретная обработка предсказания является такой же, как на этапе 322 с Фиг. 28 в примере 3.[0185] When calculating the pitch lag, the pitch lag prediction unit 312 calculates a predicted value

lagging the pitch by using the lagging pitch obtained from the audio decoding unit (step 341 in FIG. 30). The specific prediction processing is the same as in step 322 of FIG. 28 in example 3.

запаздывания основного тона и значения прошлого запаздывания

основного тона, согласно следующим уравнениям (этап 342 на Фиг. 30).[0186] Then, a pitch lag codebook is generated from the predicted value

pitch lags and past lag values

pitch according to the following equations (block 342 in FIG. 30).

><When

>

Уравнение 40

Equation 40

><When

>

Уравнение 41

Equation 41

. Дополнительно, число индексов кодовой книги равно I.

является некоторой предопределенной дискретой и

является некоторой предопределенной константой.The procedure is the same as step 331 in FIG. 29. The value of the pitch lag one subframe earlier is

... Additionally, the number of codebook indices is I.

is some predefined discrete and

is some predefined constant.

основного тона, соответствующее индексу idx, передаваемому в качестве части побочной информации, вычисляется и сохраняется в блоке 126 сбора побочной информации (этап 343 на Фиг. 30).[0187] Then, referring to the pitch lag codebook, the lag

the pitch corresponding to the index idx transmitted as the side information part is calculated and stored in the side information collecting unit 126 (step 343 in FIG. 30).

[0188] <When packet loss is detected>

Although the functional configuration of the audio synthesis block is also the same as in example 1 (which is the same as in FIG. 15), this document describes only the adaptive codebook calculator 1123 hereinafter, which operates differently from the block in the example. one.

[0189] The missing audio parameter processing unit 123 reads the pitch lag from the side information collection unit 126 and calculates the predicted pitch lag value according to the following equation, and uses the calculated predicted pitch lag value instead of outputting the pitch lag prediction unit 192.

Уравнение 42

Equation 42

является некоторой предопределенной константой.Where

is some predefined constant.

запаздывания основного тона исходный вектор

возбуждения генерируется согласно следующему уравнению (этап 332 на Фиг. 29).[0190] Then, by using the adaptive codebook and the predicted value

pitch lag original vector

excitation is generated according to the following equation (step 332 in FIG. 29).

Уравнение 43

Equation 43

основного тона, чтобы посредством этого сгенерировать вектор u(n) адаптивной кодовой книги. Для гортанной импульсной синхронизации используется такая же процедура, как на этапе 333 с Фиг. 29.[0191] Then the laryngeal pulse sync is applied to the original excitation vector by using a lag

pitch to thereby generate an adaptive codebook vector u (n). For laryngeal pulse synchronization, the same procedure as in step 333 of FIG. 29.

[0192] Hereinafter, an audio encoding program 70 that causes a computer to execute the above-described processing by an audio signal transmission apparatus is described. As shown in FIG. 31, an audio encoding program 70 is stored in a program storage area 61 formed on a recording medium 60 that is inserted into a computer and is available to or included in a computer.

[0193] The audio coding program 70 includes an audio coding unit 700 and a side information coding unit 701. The functions implemented by executing the audio coding unit 700 and the side information coding unit 701 are the same as the functions of the audio coding unit 111 and the side information coding unit 112 in the audio signal transmission apparatus described above, respectively.

[0194] It should be noted that part or all of the audio coding program 70 may be transmitted via a transmission medium such as a communication link, received and stored (including being installed) by another device. Additionally, each module of the audio encoding program 70 may not be installed on one computer, but on any of a plurality of computers. In this case, the above-described processing of the audio encoding program 70 is performed by a computer system composed of said plurality of computers.

[0195] Hereinafter, an audio decoding program 90 is described that causes a computer to execute the above-described processing by an audio signal receiving apparatus. As shown in FIG. 32, the audio decoding program 90 is stored in a program storage area 81 formed on a recording medium 80 that is inserted into a computer and is available to or included in a computer.

[0196] The audio decoding program 90 includes an audio code buffer unit 900, an audio parameter decoding unit 901, a side information decoding unit 902, a side information collection unit 903, a missing audio parameter processing unit 904, and an audio synthesis unit 905. The functions implemented by the execution of the audio code buffer unit 900, the audio parameter decoding unit 901, the side information decoding unit 902, the side information collection unit 903, the missing audio parameters processing unit 904, and the audio synthesis unit 905 are the same as the function of the audio code buffer 231, respectively. an audio parameter decoding unit 232, a side information decoding unit 235, a side information collecting unit 236, a missing audio parameter processing unit 233, and an audio synthesis unit 234 described above.

[0197] It should be noted that part or all of the audio decoding program 90 may be transmitted via a transmission medium such as a communication link, received and stored (including being installed) by another device. Additionally, each module of the audio decoding program 90 may not be installed on one computer, but on any of a plurality of computers. In this case, the above-described processing of the audio decoding program 90 is performed by a computer system composed of said plurality of computers.

[0198] [Example 4]

An example that uses side information for decoding-side pitch lag prediction is described later in this document.

[0199] <Coding side>

The functional configuration of the audio signal transmission apparatus is the same as in Example 1. The functional configuration and procedure differ only in the side information encoding section 112, and therefore, only the operation of the side information encoding section 112 will be described later in this document.

[0200] The functional configuration of an example of the side information encoding unit 112 is shown in FIG. 33 and an exemplary procedure of the side information encoding unit 112 is shown in FIG. 34. Side information coding section 112 includes an LP coefficient calculating section 511, a residual signal calculating section 512, a pitch lag calculating section 513, an adaptive codebook calculating section 514, an adaptive codebook buffer 515, and a pitch lagging encoding section 516.

[0201] The LP coefficient calculating unit 511 is the same as the LP coefficient calculating unit 151 in Example 1 shown in FIG. 8, and thus will not be described again.

[0202] The residual signal calculator 512 calculates the residual signal by the same processing as in step 181 in Example 1 shown in FIG. eleven.

, которое максимизирует следующее уравнение (этап 163 на Фиг. 34). Следует отметить, что u(n) указывает адаптивную кодовую книгу и

указывает число отсчетов, содержащихся в одном подкадре.[0203] The pitch lag calculating unit 513 calculates the pitch lag for each subframe by calculating

that maximizes the following equation (block 163 in FIG. 34). It should be noted that u (n) indicates an adaptive codebook and

indicates the number of samples contained in one subframe.

Уравнение 43

Equation 43

адаптивной кодовой книги из запаздывания

основного тона и адаптивной кодовой книги u(n). Длина адаптивной кодовой книги равна

(этап 164 на Фиг. 34).[0204] The adaptive codebook calculator 514 calculates the vector

adaptive codebook from lag

pitch and adaptive codebook u (n). The length of the adaptive codebook is

(step 164 in FIG. 34).

Уравнение 44

Equation 44

адаптивной кодовой книги (этап 166 на Фиг. 34).[0205] The adaptive codebook buffer 515 updates the state with a vector

adaptive codebook (block 166 in FIG. 34).

Уравнение 45

Equation 45

Уравнение 46

Equation 46

[0206] The pitch lag coding unit 516 is the same as in Example 1 and thus is not described again (step 169 in FIG. 34).

[0207] <Decode side>

The audio signal receiving apparatus includes an audio code buffer 121, an audio parameter decoding unit 122, a missing audio parameter processing unit 123, an audio synthesis unit 124, a side information decoding unit 125, and a side information collecting unit 126, just as in Example 1. The procedure of the audio signal receiving apparatus is shown in FIG. 7.

[0208] The operation of the audio code buffer 121 is the same as in Example 1.

[0209] <When packet is received correctly>

The operation of the audio parameter decoding unit 122 is the same as in Example 1.

основного тона и сохраняет его в блоке 126 сбора побочной информации. Блок 125 декодирования побочной информации декодирует код побочной информации посредством использования способа декодирования, соответствующего способу кодирования, используемому на стороне кодирования.[0210] The side information decoding unit 125 decodes the side information code, calculates the latency

the pitch and stores it in the block 126 collecting side information. The side information decoding unit 125 decodes the side information code by using a decoding method corresponding to the coding method used on the coding side.

[0211] The audio synthesizer 124 is the same as in example 1.

[0212] <When packet loss detected>

The ISF predictor 191 of the missing audio parameter processing unit 123 (see FIG. 12) calculates the ISF parameter in the same manner as in Example 1.

основного тона таким же способом, как в примере 1 (этап 4051 на Фиг. 35). Дополнительно, блок 192 предсказания запаздывания основного тона выводит запаздывание

основного тона посредством использования запаздывания

основного тона, использованного при прошлом декодировании (этап 4052 на Фиг. 35). Число подкадров, содержащихся в одном кадре, равно M и число запаздываний основного тона, содержащихся в побочной информации, равно

. В предсказании запаздывания

основного тона, может быть использована процедура, которая описана в непатентной литературе 4 (этап 1102 на Фиг. 13).[0213] An exemplary procedure for pitch lag prediction block 192 is shown in FIG. 35. Pitch lag prediction unit 192 reads the side information code from side information collector 126 and obtains the delay

the pitch in the same manner as in example 1 (step 4051 in FIG. 35). Additionally, the pitch lag prediction unit 192 outputs a lag

pitch by using lag

the pitch used in past decoding (step 4052 in FIG. 35). The number of subframes contained in one frame is M and the number of pitch lags contained in the side information is

... In lag prediction

pitch, the procedure described in non-patent literature 4 (step 1102 in FIG. 13) can be used.

основного тона, блок 192 предсказания запаздывания основного тона может предсказывать запаздывание

основного тона посредством использования запаздывания

основного тона, использованного при прошлом декодировании, и запаздывания

основного тона. Дополнительно,

может быть установлено. Процедура блока предсказания запаздывания основного тона в этом случае является такой, как показана на Фиг. 36.[0214] In lag prediction

pitch, pitch lag prediction block 192 can predict the pitch

pitch by using lag

the pitch used in the past decoding and the lag

the main tone. Additionally,

can be installed. The procedure of the pitch lag prediction unit in this case is as shown in FIG. 36.

только когда достоверность предсказанного значения запаздывания основного тона является низкой. Процедура блока предсказания запаздывания основного тона в этом случае показана на Фиг. 37. Информация инструкции относительно того, используется ли предсказанное значение или используется запаздывание

основного тона, полученное побочной информацией, может быть введена в блок 154 вычисления адаптивной кодовой книги.[0215] Additionally, the pitch lag prediction unit 192 may set

only when the reliability of the predicted pitch lag value is low. The procedure of the pitch lag prediction unit in this case is shown in FIG. 37. Instruction information as to whether a predicted value is used or a lag is used

the pitch obtained by the side information may be input to the adaptive codebook calculator 154.

[0216] The adaptive codebook gain predictor 193 and the fixed codebook gain predictor 194 are the same as in Example 1.

[0217] Block 195 generating noise signals is the same as in example 1.

[0218] The audio synthesis unit 124 synthesizes, from the parameters outputted from the missing audio parameter processing unit 123, an audio signal corresponding to the frame to be encoded.

[0219] The LP coefficient calculating unit 1121 of the audio synthesis unit 124 (see Fig. 15) obtains the LP coefficient in the same manner as in Example 1 (step S11301 in Fig. 16).

.[0220] The adaptive codebook calculator 1123 calculates the adaptive codebook vector in the same manner as in Example 1. The adaptive codebook calculator 1123 may or may not perform filtering on the adaptive codebook vector. In particular, the adaptive codebook vector is computed using the following equation. The filtration coefficient is

...

Уравнение 47

Equation 47

(этап A вычисления адаптивной кодовой книги).In the case of decoding, a value that does not indicate filtering is set

(step A of calculating the adaptive codebook).

[0221] The adaptive codebook calculator 1123 may calculate the adaptive codebook vector in the following procedure (step B of calculating the adaptive codebook).

[0222] The original adaptive codebook vector is computed using pitch lag and adaptive codebook 1122.

Уравнение 48

Equation 48

может быть установлено согласно методике расчета.

can be set according to the calculation method.

в варианте осуществления настоящего изобретения, и последний достоверный основной тон

соответствует

в варианте осуществления настоящего изобретения.[0223] Then, guttural pulse sync is applied to the original adaptive codebook vector. For laryngeal pulse synchronization, the same procedure is used as in the case where pulse position is not available in section 7.11.2.5 in non-patent literature 4. It should be noted, however, that u (n) in non-patent literature 4 corresponds to v (n) in variant implementation of the present invention, and the extrapolated pitch corresponds to

in an embodiment of the present invention, and the last valid pitch

corresponds

in an embodiment of the present invention.

[0224] Additionally, in the case where the pitch lag prediction unit 192 outputs the above-described instruction information for the predicted value, when the instruction information indicates that the pitch lag transmitted as side information should not be used as the predicted value (No on 38), the adaptive codebook calculator 1123 may use the adaptive codebook calculating step A above, and if it is indicated that a pitch value is to be used (Yes at 4082 in FIG. 38), the adaptive calculator 1123 the codebook can use the above-described step B of calculating the adaptive codebook. The procedure of the adaptive codebook calculator 1123 in this case is shown in the example of FIG. 38.

[0225] The excitation vector synthesis unit 1124 outputs the excitation vector in the same manner as in Example 1 (step 11306 in FIG. 16).

[0226] Post-filter 1125 performs post-processing on the synthesis signal in the same manner as in example 1.

[0227] The adaptive codebook 1122 updates the state by using the excitation signal vector in the same manner as in example 1 (step 11308 in FIG. 16).

[0228] The synthesis filter 1126 synthesizes the decoded signal in the same manner as in example 1 (step 11309 in FIG. 16).

[0229] The perceptual inverse weighting filter 1127 applies a perceptual inverse weighting filter in the same manner as in Example 1.

[0230] The missing audio parameter processing unit 123 stores the audio parameters (ISF parameter, pitch lag, adaptive codebook gain, fixed codebook gain) used in the audio synthesizer 124 into a buffer in the same manner as in Example 1 (step 145 on Fig. 7).

[0231] [Example 5]

In this embodiment, a configuration is described in which the pitch lag is transmitted as side information only in some specific class of frames, and otherwise the pitch lag is not transmitted.

[0232] <Transfer side>

In the audio signal transmission apparatus, the input audio signal is sent to the audio coding unit 111.

[0233] The audio coding unit 111 in this example calculates an index representing characteristics of a frame to be encoded and transmits the index to the side information coding unit 112. Other operations are the same as in example 1.

[0234] In the side information coding unit 112, the difference from Examples 1 to 4 exists only with respect to the pitch lag coding unit 158, and therefore the operation of the pitch lag coding unit 158 is described later in this document. The configuration of the side information encoding unit 112 in Example 5 is shown in FIG. 39.

[0235] The procedure of the pitch lag coding unit 158 is shown in the example of FIG. 40. Pitch lag coding unit 158 reads out an index representing characteristics of the frame to be encoded (block 5021 in FIG. 40), and when the index representing characteristics of the frame to be encoded is equal to some predetermined value, lag coding unit 158 pitch specifies the number of bits to be assigned to side information, as B bits (B> 1). On the other hand, when an index representing characteristics of a frame to be encoded differs from some predetermined value, pitch lag coding unit 158 determines the number of bits to be assigned to side information as 1 bit (block 5022 in FIG. 40 ).

[0236] When the number of bits to be assigned to the side information is 1 bit (No in step 5022 in FIG. 40), a value indicating that there is no side information transmission is used as the side information code and is set to the side information index. information (block 5023 in Fig. 40).

[0237] On the other hand, when the number of bits to be assigned to the side information is B bits (Yes in step 5022 in FIG. 40), a value indicating transmission of the side information is set to the side information index (step 5024 in FIG. 40), and additionally, a B-1 bit code obtained by encoding the pitch lag in the manner described in Example 1 is added for use as the side information code (step 5025 in FIG. 40).

[0238] <Decode side>

The audio signal receiving apparatus includes an audio code buffer 121, an audio parameter decoding unit 122, a missing audio parameter processing unit 123, an audio synthesis unit 124, a side information decoding unit 125, and a side information collecting unit 126, just as in Example 1. The procedure of the audio signal receiving apparatus is as shown in FIG. 7.

[0239] The operation of the audio code buffer 121 is as in Example 1.

[0240] <When packet is received correctly>

The operation of the audio parameter decoding unit 122 is the same as in Example 1.

[0241] The procedure of the side information decoding unit 125 is shown in the example of FIG. 41. The side information decoding section 125 first decodes the side information index contained in the side information code (block 5031 in FIG. 41). When the side information index indicates that there is no side information transmission, the side information decoding section 125 does not perform any additional decoding operations. Also, the side information decoding section 125 stores the side information index value in the side information collecting section 126 (step 5032 in FIG. 41).

основного тона и сохраняет вычисленное запаздывание основного тона в блоке 126 сбора побочной информации (этап 5033 на Фиг. 41). Дополнительно, блок 125 декодирования побочной информации сохраняет значение индекса побочной информации в блок 126 сбора побочной информации. Следует отметить, что декодирование побочной информации B-1 битов является такой же операцией, как операция блока 125 декодирования побочной информации в примере 1.[0242] On the other hand, when the side information index indicates transmission of side information, the side information decoding section 125 further performs decoding of B-1 bits and calculates the latency

and stores the calculated pitch lag in the side information collection unit 126 (step 5033 in FIG. 41). Additionally, the side information decoding section 125 stores the side information index value in the side information collecting section 126. It should be noted that the decoding of the B-1 bit side information is the same operation as that of the side information decoding unit 125 in Example 1.

[0243] The audio synthesizer 124 is the same as in example 1.

[0244] <When Packet Loss Detected>

The ISF predictor 191 of the missing audio parameter processing unit 123 (see FIG. 12) calculates the ISF parameter in the same manner as in Example 1.

[0245] The procedure of the pitch lag prediction block 192 is shown in the example of FIG. 42. Pitch lag prediction unit 192 reads the side information index from side information collecting unit 126 (step 5041 in FIG. 42) and checks if it is a value indicating transmission of side information (step 5042 in FIG. 42).

[0246] <When the side information index is a value indicating transmission of side information>

основного тона (этап 5043 на Фиг. 42). Дополнительно, запаздывание

основного тона выводится посредством использования запаздывания

основного тона, использованного при прошлом декодировании, и

, полученного в качестве побочной информации (этап 5044 на Фиг. 42). Число подкадров, содержащихся в одном кадре, равно M и число запаздываний основного тона, содержащихся в побочной информации, равно

. В предсказании запаздывания

основного тона, может быть использована процедура, которая описана в непатентной литературе 4 (этап 1102 на Фиг. 13). Дополнительно, может быть установлено

.In the same way as in example 1, the side information code is read from the side information collecting unit 126 to obtain the delay.

pitch (block 5043 in FIG. 42). Additionally, lag

the pitch is output by using a lag

the pitch used in past decoding, and

received as side information (step 5044 in FIG. 42). The number of subframes contained in one frame is M and the number of pitch lags contained in the side information is

... In lag prediction

pitch, the procedure described in non-patent literature 4 (step 1102 in FIG. 13) can be used. Additionally, can be installed

...

только, когда достоверность предсказанного значения запаздывания основного тона является низкой, и иначе устанавливать предсказанное значение в

(этап 5046 на Фиг. 42). Дополнительно, информация инструкции запаздывания основного тона, указывающая, используется ли предсказанное значение или используется запаздывание

основного тона, полученное побочной информацией, может быть введена в блок 1123 вычисления адаптивной кодовой книги.[0247] Additionally, the pitch lag prediction unit 192 may set

only when the confidence of the predicted pitch lag value is low, and otherwise set the predicted value to

(block 5046 in FIG. 42). Additionally, pitch lag instruction information indicating whether a predicted value is used or a lag is used

the pitch obtained by the side information may be input to the adaptive codebook calculator 1123.

[0248] <When the side information index is a value indicating that there is no side information transmission>

основного тона, блок 192 предсказания запаздывания основного тона предсказывает запаздывание

основного тона посредством использования запаздывания

основного тона, использованного при прошлом декодировании (этап 5048 на Фиг. 42).In lag prediction

pitch, pitch lag prediction block 192 predicts lag

pitch by using lag

the pitch used in past decoding (block 5048 in FIG. 42).

только когда достоверность предсказанного значения запаздывания основного тона является низкой (этап 5049 на Фиг. 42), и блок 192 предсказания запаздывания основного тона может иначе установить предсказанное значение в

. Дополнительно, информация инструкции запаздывания основного тона, указывающая, используется ли предсказанное значение или используется запаздывание

основного тона, использованное при прошлом декодировании, вводится в блок 1123 вычисления адаптивной кодовой книги (этап 5050 на Фиг. 42).[0249] Additionally, the pitch lag prediction unit 192 may set

only when the confidence of the predicted pitch lag value is low (block 5049 in FIG. 42) and the pitch lag prediction unit 192 can otherwise set the predicted value to

... Additionally, pitch lag instruction information indicating whether a predicted value is used or a lag is used

the pitch used in past decoding is input to adaptive codebook calculator 1123 (block 5050 in FIG. 42).

[0250] The adaptive codebook gain predictor 193 and the fixed codebook gain predictor 194 are the same as in Example 1.

[0251] Block 195 generating noise signals is the same as in example 1.

[0252] The audio synthesizer 124 synthesizes, from the parameters outputted from the audio missing parameter processing unit 123, an audio signal that corresponds to a frame to be encoded.

[0253] The LP coefficient calculating unit 1121 of the audio synthesizing unit 124 (see FIG. 15) obtains the LP coefficient in the same manner as in Example 1 (step S11301 in FIG. 16).

.[0254] The procedure of the adaptive codebook calculator 1123 is shown in the example of FIG. 43. The adaptive codebook calculator 1123 calculates the adaptive codebook vector in the same manner as in Example 1. First, by referring to the pitch lag instruction information (step 5051 in FIG. 43) when the confidence of the predicted value is low (Yes on step 5052 in Fig. 43), the adaptive codebook vector is calculated using the following equation (step 5055 in Fig. 43). The filtration coefficient is

...

Уравнение 49

Equation 49

может быть установлено согласно методике расчета.It should be noted that

can be set according to the calculation method.

[0255] By referring to the pitch lag instruction information when the confidence of the predicted value is high (No at step 5052 in FIG. 43), the adaptive codebook calculator 1123 calculates the adaptive codebook vector by the following procedure.

[0256] First, an original adaptive codebook vector is calculated using pitch lag and adaptive codebook 1122 (block 5053 in FIG. 43).

Уравнение 50

Equation 50

может быть установлено согласно методике расчета.

can be set according to the calculation method.

в варианте осуществления настоящего изобретения, и последний достоверный основной тон

соответствует

в варианте осуществления настоящего изобретения.[0257] Then, guttural impulse sync is applied to the original adaptive codebook vector. For laryngeal pulse synchronization, the same procedure can be used as in the case where the pulse position is not available in section 7.11.2.5 in Non-Patent Literature 4 (step 5054 in FIG. 43). It should be noted that however, u (n) in Non-Patent Literature 4 corresponds to v (n) in an embodiment of the present invention, the extrapolated pitch corresponds to

in an embodiment of the present invention, and the last valid pitch

corresponds

in an embodiment of the present invention.

[0258] The excitation vector synthesizing unit 1124 outputs the excitation signal vector in the same manner as in Example 1 (step 11306 in FIG. 16).

[0259] Post-filter 1125 performs post-processing on the synthesis signal in the same manner as in example 1.

[0260] The adaptive codebook 1122 updates the state using the excitation signal vector in the same manner as in example 1 (step 11308 in FIG. 16).

[0261] The synthesis filter 1126 synthesizes the decoded signal in the same manner as in example 1 (step 11309 in FIG. 16).

[0262] The perceptual inverse weighting filter 1127 applies a perceptual inverse weighting filter in the same manner as in Example 1.

[0263] The missing audio parameter processing unit 123 stores the audio parameters (ISF parameter, pitch lag, adaptive codebook gain, fixed codebook gain) used in the audio synthesizer 124 to a buffer in the same manner as in Example 1 (step 145 on Fig. 7).

Reference designator list

[0264] 60,80 ... data carrier, 61, 81 ... program storage area, 70 ... audio coding program, 90 ... audio decoding program, 111 ... audio coding unit, 112 ... side information coding unit, 121, 231 ... audio code buffer, 122, 232 ... audio parameters decoding unit, 123, 233 ... processing unit for missing audio parameters, 124, 234 ... audio synthesis unit, 125, 235 .. . side information decoding unit, 126, 236 ... side information collecting unit, 151, 511, 1121 ... LP coefficient calculating unit, 152, 2012 ... target signal calculating unit, 153, 513, 2013 ... unit pitch lag computation, 154, 1123, 514, 2014, 2313 ... adaptive codebook computation unit, 155, 1124, 2314 ... excitation vector synthesis unit, 156, 315, 515, 2019 ... adaptive codebook buffer , 157, 1126, 2018, 2316 ... synthesis filter, 158, 516 ... pitch lag coding block, 191 ... ISF prediction block, 192 ... pitch lag prediction block, 193 ... prediction block adaptive codebook gain knowledge, 194 ... fixed codebook gain prediction block, 195 ... noise generation block, 211 ... main coding block, 212 ... side information coding block, 213, 238 ... block masking signal collection, 214 ... error signal coding unit, 237 ... error signal decoding unit, 311 ... LP coefficient calculation unit, 312 ... pitch lag prediction unit, 313 ... pitch lag selection unit , 314 ... pitch lag coding unit, 512 ... residual signal calculator, 700 ... audio coding unit, 701 ... side information coding unit, 900 ... audio parameter decoding unit, 901 ... unit processing missing audio parameters, 902 ... audio synthesis module, 903 ... side information decoding module, 1128 ... side information output determination unit, 1122, 2312 ... adaptive codebook, 1125 ... post-filter, 1127 .. .perceptual inverse weighting filter, 2011 .. ISF coding unit, 2015 ... fixed codebook computation unit, 2016 ... gain calculation unit, 2017 ... excitation vector computation unit, 2211 ... ISF decoding unit, 2212 ... pitch lag decoding unit , 2213 ... gain decoding unit, 2214 ... fixed codebook decoding unit, 2318 ... synthesis unit for predictive excitation vectors.

Claims (11)

1. An audio coding method by an audio coding device for coding an audio signal, comprising: an audio encoding step that encodes an audio signal; a side information encoding step in which side information is calculated from the prediction signal to calculate the predicted value of an audio parameter in order to synthesize the encoded audio, and the side information is encoded, wherein the side information contains information indicating the availability of side information, and the side information is taken as an audio parameter on the decoding processing side when the confidence of the predicted value of the computed audio parameter is low. 2. The audio coding method of claim 1, wherein the side information indicates a pitch lag included in the prediction signal. 3. An audio encoding device for encoding an audio signal, said audio encoding device comprising: an audio encoder configured to encode an audio signal; and a side information encoder configured to compute side information from a predicted signal to calculate a predicted value of an audio parameter to synthesize encoded audio, and encode the side information, wherein the side information contains information indicating the availability of side information, and the side information is received as an audio parameter on the decoding processing side when the reliability of the predicted value of the computed audio parameter is low.

Images